STEM CELLS AND DEVELOPMENT Volume 25, Number 6, 2016 Mary Ann Liebert, Inc. DOI: 10.1089/scd.2015.0298

Profiling the Secretome of Human Stem Cells from Dental Apical Papilla

Shi Yu,1 Yuming Zhao,1 Yushi Ma,2 and Lihong Ge1

Recent studies have shown that secretion of bioactive factors from mesenchymal stem cells (MSCs) plays a primary role in MSC-mediated therapy; especially for bone marrow-derived MSCs (BMSCs). MSCs from dental apical papilla (SCAPs) are involved in root development and may play a critical role in the formation of dentin and pulp. Bioactive factors secreted from SCAPs actively contribute to their environment; however, the SCAPs secretome remains unclear. To address this and gain a deeper understanding of the relevance of SCAPs secretions in a clinical setting, we used isobaric chemical tags and high-performance liquid chromatography with tandem mass spectrometry to profile the secretome of human SCAPs and then compared it to that of BMSCs. A total of 2,046 were detected from the conditioned medium of SCAPs, with a false discovery rate of less than 1.0%. Included were chemokines along with angiogenic, immunomodulatory, antiapoptotic, and neuroprotective factors and extracellular matrix (ECM) proteins. The secreted levels of 151 proteins were found to differ by at least twofold when BMSCs and SCAPs were compared. Relative to BMSCs, SCAPs exhibited increased secretion of proteins that are involved in metabolic processes and transcription and lower levels of those associated with biological adhesion, developmental processes, and immune function. In addition, SCAPs secreted significantly larger amounts of chemokines and neurotrophins than BMSCs, whereas BMSCs

secreted more ECM proteins and proangiogenic factors. These results may provide important clues regarding the molecular mechanisms associated with tissue regeneration and how they differ between cell sources.

Introduction ment [10], and in combination with tissue engineering pro- cedures, Huang et al. used SCAPs to regenerate vascularized he secretions of mesenchymal stem cells (MSCs) human dental pulp in an emptied root canal and this was Thave gained considerable attention in recent years, par- accompanied by formation of new dentin on the existing ticularly with regard to MSC-mediated therapy and tissue dentinal wall [11]. However, the understanding of SCAPs regeneration. This has resulted in a large body of evidence function and relevance in a clinical setting needs to be im- suggesting that differentiation of MSCs into multiple cell proved. One means to achieve this is identification of the types is limited under in vivo conditions and the secretion of wide array of proteins they secrete. bioactive factors from the MSCs themselves plays a critical Therefore, in this study, we investigated the secretome of role [1–3]. MSCs actively contribute to their environment by human SCAPs and compared the resulting profile to that of releasing trophic (antiapoptotic, supportive, angiogenic, etc.) BMSCs, with the aim of achieving insights into the func- factors, immunomodulatory factors, and chemoattractants tional role of SCAPs secretions and the molecular basis of that act either on themselves or on neighboring cells [4]. their actions. Various studies have shown that the use of bone marrow- derived MSC (BMSC)-derived conditioned media (CM) alone without cell transplantation can induce tissue repair, Materials and Methods including that of bone and nerves [5–7]. Establishment, characterization, and analysis Stem cells from dental apical papilla (SCAPs) are found of cell proliferation rate of SCAPs in the root apex of immature permanent teeth [8]. These and BMSCs cultures cells appear to be the source of odontoblasts and contribute to root formation [9]. For example, interruption of SCAPs Normal human impacted third molars with immature roots development was found to result in a halted root develop- were collected from healthy patients (aged 16–24 years)

1Department of Pediatric Dentistry, Peking University School and Hospital of Stomatology, Peking University Health Science Center, Peking University, Beijing, China. 2The Department of Orthodontics Center for Craniofacial Research and Regeneration, Beijing, China.

499 500 YU ET AL.

(n = 5). Bone marrows were obtained from the mandibular (release 2014_04_10; 20,264 reviewed entries) was used to bone of healthy patients (aged 20–30 years) (n = 3). Sample search. A fragment-ion mass tolerance of 0.050 Da and a collection was approved by the Ethics Committee of the parent-ion tolerance of 10.0 ppm were equipped for Mascot. Health Science Center of Peking University (Beijing, China; Scaffold Q+ version 4.4.1.1 (Proteome Software, Inc.) IRB00001052-11060). SCAPs and BMSCs were isolated was used to assess label-based peptide quantification and from dental apical papilla tissues and mandibular bone mar- identification. Peptide identifications were accepted row, respectively. Cultures of MSCs were maintained in if the false discovery rate (FDR) was less than 1.0%, and alpha-modified Eagle’s minimum essential medium (Gibco) protein identifications were accepted if the probability was supplemented with 10% fetal bovine serum (Gibco) in 5% more than 66.0% and at least one identified peptide was carbon dioxide at 37C. The cells were used in experiments contained. Acquired intensities in the experiment were after three to five passages, and for each experiment, SCAPs globally normalized across all acquisition runs. Individual and BMSCs had the same passage number. quantitative samples were normalized within each acquisi- SCAPs and BMSCs were characterized by flow cytometry tion run. The intensity of each peptide identified was nor- using fluorescein-isothiocyanate-conjugated or phycoerythrin- malized within the assigned protein. The reference channels conjugated antibodies specific for CD73, CD90, CD105, were normalized to produce a 1:1 ratio. All normalization CD146, and CD34 (BD Biosciences) as previously described. calculations were performed using medians to multiplica- MSCs were seeded into 96-well plates at a density of 1.0 · tively normalize the data. Differentially expressed proteins 103 cells/well and then cultured for 7 days. Cell counting kit were determined using a Kruskal–Wallis test. String soft- 8 (CCK-8; Dojindo) solution was added to each well of the ware (http://string-db.org/) was employed to visualize pro- plate and the absorbance was measured at 450 nm, every 24 h, tein interaction networks between the secretory proteins of according to the manufacturer’s protocol. SCAPs. Functional annotation analysis using ontology (GO) terms was performed using the DAVID Bioinfor- Preparation of CM and protein digestion matics Resources 6.7 (http://david.abcc.ncifcrf.gov/) and PANTHER (protein analysis through evolutionary relation- SCAPs and BMSCs were seeded on 100-mm plates at a ships) classification system (www.pantherdb.org/). density of 20,000 cells/cm2. When MSCs reached 90% confluency, they were washed five times with phosphate- buffered saline and cultured in serum-free medium for 24 h. Western blotting The CM were collected after centrifugation at 130 g for Thirty micrograms of secretory protein were separated 10 min to remove the cellular debris and then passed through by 10% sodium dodecyl sulfate–polyacrylamide gel elec- a 0.22-mm filter. The samples were concentrated, then air-

trophoresis and transferred on a polyvinylidene fluoride dried, redissolved in triethylammonium bicarbonate (Pro- membrane. The membrane was blocked with 5% (w/v) nonfat mega), and finally, reduced with dithiothreitol at 55C for dried milk, incubated overnight at 4C with primary antibodies 1 h. Next, iodoacetamide (Promega) was added to the sam- against midkine (MDK) and angiopoietin-related protein 2 ples and they were left for 1 h at room temperature, pro- (ANGPTL2; Abcam), and then reacted with horseradish- tected from light. Protein concentration was determined peroxidase-conjugated secondary antibodies (Origene). using a bicinchoninic acid (Thermo Fisher Scientific) assay. Immunoreactive bands were visualized by enhanced chemi- luminescence (Cwbiotech) at room temperature and then TMT and HPLC-MS/MS analysis digitized using the Fusion FX image analyzer (Viber Loumat). Samples were digested with sequence-grade modified trypsin (Promega) and then labeled using the TMT reagent kit Real-time reverse transcription-polymerase (Thermo Fisher Scientific) [12]. Ten fractions were collected chain reaction analysis by high-pH separation using an Aquity UPLC system (Waters Corporation) [13]. Total RNA was extracted from SCAPs and BMSCs (after The fractions were resuspended in formic acid, separated the culture procedure described above) using TRIzol Reagent by nanoLC, and analyzed by online electrospray tandem mass (Invitrogen) according to the manufacturer’s instructions. spectrometry. The analysis was performed with nanoAquity cDNA was synthesized with oligo(dT) primers using a re- UPLC system (Waters Corporation), which was equipped verse transcriptase kit (Promega). The primer sequences were with an online nanoelectrospray ion source and connected to designed by primer bank (Supplementary Table S1; Supple- a Q-Exactive hybrid quadrupole-orbitrap mass spectrometer mentary Data are available online at www.liebertpub.com/ (Thermo Fisher Scientific). Samples were separated by ana- scd). Quantitative reverse transcription polymerase chain re- lytical column (Thermo Scientific Acclaim PepMap C18, action (qRT-PCR) was carried out in triplicate in 96-well 75 mm · 50 cm) with a linear gradient from 5% D to 30% D plates using a 7900HT Fast Real-Time system (Applied -DDCT in 95 min. Full-scan MS spectra (m/z 350–1200) were ac- Biosystems). The comparative cycle threshold (2 ) quired and three MS/MS events were analyzed. The samples method was used to calculate the relative expression levels of were analyzed in triplicate. the target . GAPDH was used as an internal control.

Database searching and quantitative data analysis Statistical analysis MS was extracted by ProteoWizard version 3.0.5126 Statistical analysis was performed using SPSS version (Thermo Fisher Scientific) and analyzed using Mascot version 15.0 (SPSS). Student’s t-tests were used to assess the sig- 2.3 (Matrix Science). The human Uniprot-SwissProt database nificance of differences between groups; a P value of <0.05 PROFILING THE SECRETOME OF SCAPS 501

FIG. 1. SCAPs exhibited increased cell proliferation compared with BMSCs. (A) CCK-8 assay showed that the relative OD values for SCAPs were significantly higher (Student’s t-test; P < 0.05) than those for BMSCs 4 days after seeding (**P < 0.01). Results represent the mean – SD from four independent experiments. (B) SCAPs and BMSCs were positive for CD73, CD90, CD105, and CD146 and were negative for CD34. BMSC, bone marrow-derived mesenchymal stem cells; OD, optical density; SCAP, stem cells from dental apical papilla. was considered an indicator of statistical significance. All Cell surface-marker analysis by flow cytometry showed experiments were repeated at least thrice (n ‡ 3). that both stem cell types were positive for CD73, CD90, CD105, and CD146 and negative for CD34 (Fig. 1B). Results Proliferation of SCAPs and BMSCs Quantitative analysis of low-pH nano-HPLC-MS/MS Cell proliferation was monitored over a period of 7 days Cell death was rarely detected during the incubation pe- postseeding, and we found that SCAPs showed an enhanced riod, as determined by microscopy. A total of 2,046 proteins proliferation capacity versus BMSCs (Fig. 1A). were detected from the CM of the SCAPs, and the FDR was

FIG. 2. Functional interaction network of the ECM proteins in the SCAPs sectetome. String software was used to analyze the ECM proteins listed in Supplementary Table S2. ECM, extracellular matrix.

Table 1. Trophic and Immunomodulatory Factors Secreted by SCAPs and BMSCs Gene Accession Fold change Effect Identified proteins name no. RCoverage R. PSMs P value (SCAPs/BMSCs) Angiogenic Vascular endothelial growth factor C VEGFC P49767 37.95 44 0.0056 0.8 Platelet-derived growth factor receptor beta PDGFRB P09619 23.42 61 0.00022 0.7 Angiopoietin-related protein 1 ANGPTL1 Q15389 31.73 47 0.00021 0.8 Angiopoietin-related protein 2 ANGPTL2 Q9UKU9 24.54 47 <0.0001 0.4 Angiopoietin-related protein 4 ANGPTL4 Q9BY76 19.21 16 0.021 0.7 Angiopoietin 1 ANGPT1 Q15389 31.73 47 0.00021 0.8 Fibroblast growth factor 7 FGF7 P21781 32.99 10 0.39 0.6 Hepatocyte growth factor HGF P14210 19.09 52 0.12 0.5 72 kDa type IV collagenase MMP2 P08253 70 653 <0.0001 0.8 Matrix metalloproteinase-14 MMP14 P50281 24.4 57 0.046 0.6 Interstitial collagenase MMP1 P03956 39.87 53 0.00078 0.6 Matrix metalloproteinase 14 MMP14 P50281 24.4 57 0.0018 0.7 Matrix metalloproteinase 19 MMP19 Q99542 34.25 34 0.11 1.8 Vascular cell adhesion protein 1 VCAM1 P19320 33.42 118 <0.0001 0.5

502 Integrin beta-1 ITGB1 P05556 29.2 65 0.0039 0.6 Immunomodulatory Interleukin 6 IL6 P05231 26.42 14 0.019 0.4 Macrophage colony-stimulating factor 1 CSF1 P09603 26.9 50 <0.0001 0.6 Follistatin-related protein 1 FSTL1 Q12841 79.22 292 <0.0001 0.7 Pentraxin-related protein PTX3 PTX3 P26022 51.97 107 <0.0001 0.4 Transforming growth factor beta 1 TGFB1 P01137 24.87 50 0.4 1.1 Transforming growth factor beta 2 TGFB2 P61812 30.19 68 <0.0001 2.1 Galectin 1 LGALS1 P09382 81.48 146 <0.0001 1.4 Galectin 3 LGALS3 P17931 40 48 0.15 1.3 Macrophage migration inhibitory factor MIF P14174 27.83 19 0.019 1.5 Antiapoptotic Metalloproteinase inhibitor 1 TIMP1 P01033 57.49 210 0.00082 0.5 Metalloproteinase inhibitor 2 TIMP2 P16035 60.45 226 <0.0001 1.1 Metalloproteinase inhibitor 4 TIMP4 Q99727 19.64 13 0.012 0.2 Insulin-like growth factor-binding protein 2 IGFBP2 P18065 53.54 170 <0.0001 0.6 Insulin-like growth factor-binding protein 4 IGFBP4 P22692 62.4 278 <0.0001 0.6 Insulin-like growth factor-binding protein 5 IGFBP5 P24593 51.47 122 <0.0001 1.3 Insulin-like growth factor-binding protein 6 IGFBP6 P24592 77.08 82 <0.0001 0.6 Insulin-like growth factor-binding protein 7 IGFBP7 Q16270 69.86 268 <0.0001 0.3 Insulin-like growth factor II IGF2 P01344 61.67 69 <0.0001 1.4

(continued)

Table 1. (Continued) Gene Accession Fold change Effect Identified proteins name no. RCoverage R. PSMs P value (SCAPs/BMSCs) Hepatocyte growth factor HGF P14210 19.09 52 0.083 0.5 Heat shock 70 kDa protein 4 HSPA4 P34932 51.07 181 0.00085 1.2 Heat shock protein HSP 90-beta HSP90AB1 P08238 40.33 137 0.00057 1.8 Stress-70 protein, mitochondrail HSPA9 P38646 36.38 110 <0.0001 2.5 Chemokines Stromal cell-derived factor 1 SDF1 P48061 39.78 12 0.00026 2.3 Nurotrophin Midkine MDK P21741 62.24 45 0.00078 3.3 Pleiotrophin PTN P21246 61.31 39 0.0027 3.0 Mesencephalic astrocyte-derived MANF P55145 32.42 27 0.068 1.1 neurotrophic factor Neuroblast differentiation-associated AHNAK Q09666 74.79 1611 <0.0001 1.6 protein AHNAK Neuropilin 2 NRP2 O60462 26.96 80 0.00012 1.7 Extracellular matrix Collagen alpha 1(I) chain COL1A1 P02452 83.74 2123 <0.0001 0.8 503 Collagen alpha 2(I) chain COL1A2 P08123 86.68 1920 <0.0001 0.6 Collagen alpha 1(III) chain COL3A1 P02461 63.85 845 <0.0001 0.3 Collagen alpha 1(V) chain COL5A1 P20908 41.73 409 <0.0001 0.7 Collagen alpha 2(V) chain COL5A2 P05997 58.51 671 <0.0001 0.4 Collagen alpha 2(VI) chain COL6A2 P12110 60.26 471 <0.0001 0.3 Collagen alpha 3(VI) chain COL6A3 P12111 57.88 1138 <0.0001 0.7 Laminin subunit alpha 1 LAMA1 P25391 14.86 106 <0.0001 0.4 Laminin subunit gamma 1 LAMC1 P11047 44.38 600 <0.0001 0.6 Others Retinol-binding protein 1 RBP1 P09455 66. 67 23 0.021 3.8 Dickkopf-related protein 1 DKK1 O94907 31.95 36 0.031 1.6 Dickkopf-related protein 3 DKK3 Q9UBP4 50 183 <0.0001 1.1 Bone morphogenetic protein 1 BMP1 P13497 30.22 126 <0.0001 0.7 Mimecan OGN P20774 26.17 57 0.00042 2.8 Annexin A2 ANXA2 P07355 63.42 160 <0.0001 0.5 Thrombospondin 1 THBS1 P07996 54.96 487 <0.0001 0.5 Thrombospondin 2 THBS2 P35442 37.29 174 <0.0001 0.5 Signal transducer and activator STAT3 P40763 15.45 28 0.32 1.3 of transcription 3 Tumor suppressor p53-binding protein 1 TP53BP1 Q12888 19.47 121 0.0039 2.3

BMSC, bone marrow-derived mesenchymal stem cells; PSM, peptide-to-spectrum matches; SCAP, stem cells from dental apical papilla. 504 YU ET AL. less than 1.0%. In addition, 74.3% of the proteins (1,520/ Bioinformatic analysis of the secretome 2,046) were identified in at least two of the replicate ex- periments, including 1,172 proteins that were detected in all To further investigate the different functions of secretory three experiments. proteins between SCAPs and BMSCs, we analyzed the on- In the proteins that were identified in all three experiments, tology of the identified 151 significant different proteins 296 (25.3%) proteins matched the GO terms ‘‘extracellular (P < 0.05; fold-change ‡2 for each replicate) based on ‘‘cellular region’’ and 110 (9.5%) proteins matched the GO terms component’’ and ‘‘biological process’’ using the PANTHER ‘‘extracellular matrix (ECM)’’ using DAVID Bioinformatics classification system. After carrying out a GO analysis Resources. The extracellular region is the space external to (Fig. 3A) in the cellular-component domain, the vast majority the outermost structure of a cell, while ECM is a structure of proteins that were less secreted in SCAPs could be mapped generated and secreted by cells to provide structural and to the GO terms (categories), ‘‘extracellular matrix’’ (35%) and functional support for cells or tissues. The identified ECM ‘‘extracellular region’’ (61%). Of the 83 proteins significantly proteins contained a noticeable number of structural proteins increased in SCAPs, 36% mapped to ‘‘cell part’’ and 21% like collagens and fibronectin, as well as multiple growth mapped to ‘‘macromolecular complex’’; these two categories factors such as transforming growth factor beta 1 (TGFb1) included vimentin, ATP-dependent RNA helicase A, regula- and ANGPTL2 (Supplementary Table S2). To facilitate the tors of G-protein signaling, and eukaryotic translation initiation interpretation of the ECM proteins, a functional interaction factors. In addition, 7% of the increased proteins were ECM network of the ECM proteins was searched using the String proteins, such as stromal cell-derived factor 1 (SDF1), peri- software. As illustrated in Fig. 2, those ECM proteins se- ostin, latent TGFb-binding protein 4, and mimecan. creted by SCAPs belong to eight major functional classes: Proteins in the biological process domain are listed in cell adhesion, cell proliferation, collagen organization and Fig. 3B and of the increased proteins 45 mapped to ‘‘met- proteases, immune system, odontogenesis, angiogenesis, abolic process’’ and 33 mapped to ‘‘cellular process’’; these neuron development, and signaling molecules. two categories included 1-phosphatidylinositol 4-kinase, Then we analyzed the secretome of BMSCs (Supple- 5-bisphosphate phosphodiesterase, KH domain-containing mentary Table S5), it is noticed that the identified secretory RNA-binding signal transduction-associated protein 1 proteins from SCAPs and BMSCs, both included factors (KHDRBS1), Rho GDP-dissociation inhibitor 1, SWI/SNF involved in angiogenesis, immunomodulation, chemotaxis, complex subunit, and stress-70 protein (HSPA9). The cat- neuroprotection, antiapoptosis, and ECM formation; some egory, ‘‘developmental process’’ covered 7% of the in- of these are listed in Table 1. Most of the angiogenic factors creased proteins and decreased proteins and 9% came under and ECM proteins were significantly lower in SCAPs, when ‘‘immune system process.’’

compared to BMSCs, whereas SCAPs exhibited significantly elevated secretion of chemokines and neurotrophins. On the Validation of several differentially whole, there were no significant differences for immuno- secreted proteins modulatory and antiapoptotic factors. When comparing the secretomes of SCAPs and BMSCs, To validate our proteomic findings, we selected several we found significant differences of at least twofold in mag- proteins that were secreted in quantities that differed between nitude (P < 0.05; fold change ‡2 for each replicate) for 151 SCAPs and BMSCs, including neurotrophins [MDK; pleio- proteins. Of these, 83 were higher in SCAPs (Supplementary trophin (PTN)], development-related proteins [retinol-binding Table S3) and 68 were lower (Supplementary Table S4). protein 1 (RBP1)], an angiogenic factor (ANGPTL2),

FIG. 3. Bioinformatic analysis of proteins secreted in significantly different quantities by SCAPs (P < 0.05; ‡2-fold change vs. BMSC value). (A) Differentially secreted proteins categorized according to cellular component (inner circle: proteins decreased for SCAPs vs. BMSCs; outer circle: proteins increased for SCAPs vs. BMSCs). (B) Differentially secreted proteins categorized according to biological process. PROFILING THE SECRETOME OF SCAPS 505 antiapoptotic factors [insulin-like growth factor-binding protein duction proteins, and various transmembrane proteins [18]. 7 (IGFBP7); metalloproteinase inhibitor 4 (TIMP4)], immu- It has been found that MSCs can produce higher quantities nomodulatory factors [-related protein 3 (FSTL3)], of exosomes than other cells [19]. Several studies have re- and cell cycle regulator (KHDRBS1). The expression of the ported that MSC-derived exosomes perform functions MDK, PTN, RBP1,andKHDRBS1 genes were significantly similartothoseofMSCs,suchasrepairingtissue,sup- higher in SCAPs, as shown by qRT-PCR (Fig. 4A). Western pressing inflammatory responses, and modulating the im- blotting indicated that SCAPs had a higher secretion of MDK mune system [20,21]. proteins and a lower secretion of ANGPTL2 proteins compared We found that the proliferation capacity of SCAPs was with BMSCs (Fig. 4B). These results were consistent with our significantly greater compared with BMSCs; this may be due proteomic analysis. Also, in agreement with the proteomic re- to differences in the developmental stage of the respective cell sults was the finding that IGFBP7, TIMP4, ANGPTL2,and types. Our analysis of their secretomes showed that SCAPs FSTL3 were all significantly decreased in secreted much more proteins involved in metabolic processes, SCAPs. especially nucleic acid binding or transcription, such as KHDRBS1, Rho GDP-dissociation inhibitor 1, eukaryotic Discussion translation initiation factors, and heat shock protein (HSPs). KHDRBS1 is also called sam68 (the Src-associated substrate The spectrum of regulatory and trophic factors secreted by in mitosis of 68 kDa), which is involved in cell cycle regula- MSCs is generally referred to as the MSC secretome. To tion. It has been demonstrated that Sam68-defcient cells ex- define the role of MSC-secreted factors in tissue regeneration, hibited markedly decreased growth rate due to elongation of one should start with biomolecular profiling or secretome the G2-M phase [22]. In addition, overexpression of a Sam68 analysis of MSCs. The effect of serum must be considered in isoform with the deletion of its KH domain resulted in the MSC secretome analysis in vitro. Serum is often supple- suppression of cell growth and cyclin D1 expression, which mented to the cell culture medium for cell growth. However, could be rescued by Sam68 transfection, suggesting that the extreme complexity of proteins in serum may overlap Sam68 participates in the control of cell proliferation by pro- components of MSC secretome, which are usually secreted at moting the S phase entry [23]. Rho GDP-dissociation inhibitor low concentrations (as low as ng/mL) in the culture media 1 plays a key role in the regulation of signaling through Rho [14]. The existence of serum will interfere with protein de- GTPases. Increased expression of Rho GDP-dissociation in- tection and make their recovery difficult. To solve this hibitor 1 may have important implications in cell functions problem, MSCs can be cultured in a serum-free medium for a involving massive cytoskeletal changes such as cell prolifer- defined time, which has been successfully applied to secre- ation, differentiation, migration, and adhesion. It is reported

tome analysis in a number of studies [15–17]. Thus, we that Rho GDP-dissociation inhibitor 1 was upregulated in collect the CM under a serum-free condition in this study. dental pulp stem cells (DPSCs) compared to BMSCs [24]. Of the proteins that were increased ‡2-fold in SCAPs, HSPs function as molecular chaperones in the stabilization of 35.8% were intracellular proteins. There was little cell intracellular proteins, repairing damaged proteins, and assist- death (associated with release of cell contents) during the ing in protein translocation. It was reported that the high levels incubation period, but most of the identified proteins have of HSP90, HSC70, and HSPA9 in mouse embryonic stem cells been previously reported as present in exosomes (secreted were eliminated when the cells underwent differentiation [25]. ); thus, it is worth considering whether exo- MSCs retain the capacity to regenerate the unique micro- some formation and secretion is linked to the secretome dif- environments from which they are derived in vivo [24,26–28]. ferences we have observed. Exosomes often contain common In our study, 35% of proteins that were less secreted in cytosolic proteins, intracellular and intercellular signal trans- SCAPs were ECM proteins and only 7% of the increased

FIG. 4. Verification of proteins secreted in significantly different quantities by SCAPs and BMSCs (P < 0.05; ‡2-fold change). (A) Quantitative reverse transcription polymerase chain reaction analysis of gene expression for several proteins secreted in significantly different quantities by SCAPs versus BMSCs (**P < 0.01). Each experiment was repeated thrice. (B) Western blot showing significantly increased secretion of MDK proteins by SCAPs and significantly increased secretion of ANGPTL2 proteins by BMSCs. ANGPTL2, angiopoietin-related protein 2; MDK, midkine. 506 YU ET AL. proteins in SCAPs were ECM proteins, which is consistent induce preodontoblast differentiation and formation of func- with the function in vivo. SCAPs and BMSCs contribute to tional odontoblast-like cells in vitro [39,40]; it has also been dentin and bone formation, respectively. Actually, above shown to control mouse tooth size and development in vivo 30% of the acellular part of bone consists of the organic [41]. The overexpression of RBP1 could promote osteogenic components, and only 20% of dentin consists of the or- differentiation of BMSCs [42]. During the cap stage of crown ganic components by weight. development, RBP1-positive cells were mainly localized to the Angiogenesis, the formation of capillaries from preexisting SCAPs [43]. The secretion of RBP1 was much higher in blood vessels, is an important process in tissue engineering. SCAPs compared to BMSCs, suggesting that RBP1 would be a SCAPs have been shown to secrete proangiogenic factors and good marker for SCAPs. Choi et al. showed that knockdown of have a predominantly proangiogenic impact on endothelial SMOC1 significantly inhibited mineralization and the ex- migration and tube formation [29]. In this study, proangio- pression of osteoblast differentiation markers in BMSCs, while genic and antiangiogenic proteins were compared between overexpression of SMOC1 substantially promoted osteoblast SCAPs and BMSCs. The results indicated that secretion of differentiation [44]. The condition medium of BMSCs was proangiogenic factors was lower in SCAPs, such as vascular proved to accelerate osteogenesis in vivo and had an osteo/ endothelial growth factor C (VEGFC) and ANGPT1. VEGF odontogenic inductive effect on DPSCs in vitro [5,45,46]. It is family is the most important growth factor that regulates the speculated that the secretome of SCAPs plays an important process of angiogenesis. VEGFC is a ligand of the VEGFR-2 role in dentin regeneration and the condition medium from and VEGFR-3 receptors, which activates blood-vessel tip SCAPs may be used for the induction of dentin and bone cells [30]. ANGPT1 stimulates mural coverage and base- regeneration both in vitro and in vivo. ment membrane deposition, thereby contributing to vessel In conclusion, our study provides insights into the func- maturation and stability [31]. In addition, the secretion of tion and molecular details of SCAPs secretions, and it has homing- and transmigration-related factors, such as adhesion shown significant differences between the secretomes of molecules and matrix metalloproteinases (MMPs), was lower SCAPs and the more widely studied and utilized BMSCs. in SCAPs. The adhesion molecules, integrins and vascular SCAPs were found to secrete angiogenic, immunomodula- cell adhesion protein 1, participate in stem cell homing by tory, antiapoptotic, chemokine, and neuroprotective factors. adhesion to ECM and immunoglobulin superfamily mole- Among the most significant differences between SCAPs and cules [32–34]. MMPs are matrix-degrading enzymes that BMSCs was that SCAPs secreted considerably higher levels promote endothelial cell migration and tube formation by of chemokines and neurotrophins. These results may shed remodeling the basement membrane, by executing directional some new light on the molecular mechanisms involved in matrix proteolysis, or by exposing chemotactic cryptic motif tissue regeneration, while also demonstrating the potential

sites in the ECM [31]. Meanwhile, the antiangiogenic pro- of SCAPs as an alternative cell source for tissue engineering teins, such as TIMPs and thrombospondin, were also signif- and therapeutic applications. icantly decreased in SCAPs. TIMPs and thrombospondin function as negative regulators of angiogenesis mainly by Acknowledgments interacting with proteases and VEGF. Thus, the effect of This work was supported by grants from the National angiogenesis seemed to be more active in BMSCs compared Natural Science Foundation of China (no. 81170928 and to SCAPs. Given microenvironments the MSCs derived from, no. 81541110) and the Peking University 985 project (no. it is hypothesized that BMSCs may be more effective for 2013-4-01). We gratefully acknowledge the assistance of angiogenesis. However, further study of angiogenesis is still Wang Feng from BioyongTech for technical data analysis needed for a better understanding about the differences be- and her expertise in proteomics analysis. The authors de- tween SCAPs and BMSCs. clare no potential conflicts of interest with respect to the Our secretome profiles indicate that neurotrophins, such as authorship and/or publication of this article. MDK and PTN, were secreted in larger amounts by SCAPs when compared to BMSCs. MDK, also known as neurite Author Disclosure Statement growth-promoting factor 2 (NEGF2), is a basic heparin- binding growth factor that forms a family with PTN (also No competing financial interests exist. known as NEGF1) [35]. They are expressed in neural pre- cursor cells and promote their growth [36,37]. It was recently References reported that the secretions of BMSCs exhibited neuropro- tective effects and enhanced tissue repair in the central ner- 1. Zimmermann CE, M Gierloff, J Hedderich, Y Acil, J vous system [6]. In addition, SCAPs secreted significantly Wiltfang and H Terheyden. (2011). Survival of transplanted rat bone marrow-derived osteogenic stem cells in vivo. larger amounts of SDF1, which is a member of the CXC Tissue Eng Part A 17:1147–1156. subfamily and a widely expressed chemotactic cy- 2. Lavoie JR and M Rosu-Myles. (2013). Uncovering the se- tokine that mediates cell migration through its binding to cretes of mesenchymal stem cells. Biochimie 95:2212–2221. CXC chemokine receptor 4 (CXCR4) [38]. Our results 3. Ranganath SH, O Levy, MS Inamdar and JM Karp. (2012). showed that SCAPs secreted significantly greater quantities Harnessing the secretome for the of chemokines and neurotrophins, suggesting that SCAPs treatment of cardiovascular disease. Cell Stem Cell 10: may be good candidates for use in neural tissue repair. 244–258. Some proteins involved in odontogenesis or osteogenesis 4. Meirelles Lda S, AM Fontes, DT Covas and AI Caplan. were secreted more in SCAPs than BMSCs, such as TGFb, (2009). Mechanisms involved in the therapeutic properties RBP1, and SPARC-related modular calcium-binding protein 1 of mesenchymal stem cells. Cytokine Growth Factor Rev (SMOC1). TGFb is expressed during tooth formation and can 20:419–427. PROFILING THE SECRETOME OF SCAPS 507

5. Osugi M, W Katagiri, R Yoshimi, T Inukai, H Hibi and M secreted by MSC reduces myocardial ischemia/reperfusion Ueda. (2012). Conditioned media from mesenchymal stem injury. Stem Cell Res 4:214–222. cells enhanced bone regeneration in rat calvarial bone de- 21. Zhang B, Y Yin, RC Lai, SS Tan, AB Choo and SK Lim. fects. Tissue Eng Part A 18:1479–1489. (2014). Mesenchymal stem cells secrete immunologically 6. Tsai MJ, SK Tsai, BR Hu, DY Liou, SL Huang, MC Huang, active exosomes. Stem Cells Dev 23:1233–1244. WC Huang, H Cheng and SS Huang. (2014). Recovery of 22. Li QH, I Haga, T Shimizu, M Itoh, T Kurosaki and J Fu- neurological function of ischemic stroke by application of jisawa. (2002). Retardation of the G2-M phase progression conditioned medium of bone marrow mesenchymal stem on gene disruption of RNA binding protein Sam68 in the cells derived from normal and cerebral ischemia rats. DT40 cellline. FEBS Lett 525:145–150. J Biomed Sci 21:5. 23. Barlat I, F Maurier, M Duchesne, E Guitard, B Tocque and 7. Chen L, Y Xu, J Zhao, Z Zhang, R Yang, J Xie, X Liu and F Schweighoffer. (1997). A role for Sam68 in cell cycle S Qi. (2014). Conditioned medium from hypoxic bone progression antagonized by a spliced variant within the KH marrow-derived mesenchymal stem cells enhances wound domain. J Biol Chem 272:3129–3132. healing inmice. PLoS One 9:e96161. 24. Mrozik KM, PS Zilm, CJ Bagley, S Hack, P Hoffmann, S 8. Sonoyama W, Y Liu, T Yamaza, RS Tuan, S Wang, S Gronthos and PM Bartold. (2010). Proteomic character- Shi and GT Huang. (2008). Characterzation of the api- ization of mesenchymal stem cell-like populations derived cal papilla and its residing stem cells from human im- from ovine periodontal ligament,dental pulp, and bone mature permanent teeth: a pilot study. J Endod 34: marrow: analysis of differentially expressed proteins. Stem 166–171. Cells 19:1485–1499. 9. Huang GT, W sonoyama, Y Liu, H Liu, S Wang and S Shi. 25. Prinsloo E, MM Setati, VM Longshaw and GL Blatch. (2008). The hidden treasure in apical papilla: the potential (2009). Chaperoning stem cells: a role for heat shock pro- role in pulp/dentin regeneration and bioroot engineering. teins in the modulation of stem cell self-renewal and dif- J Endod 34:645–651. ferentiation? Bioessays 31:370–377. 10. Chueh LH and GT Huang. (2006). Immature teeth with 26. Gronthos S, M Mankani, J Brahim, PG Robey, and S Shi. periradicular periodontitis or abscess undergoing apex- (2000). Postnatal human dental pulp stem cells (DPSCs) ogenesis: a paradigm shift. J Endod 32:1205–1213. in vitro and in vivo. Proc Natl Acad Sci U S A 97: 11. Huang GT, T Yamaza, LD Shea, F Djouad, NZ Kuhn, RS 13625–13630. Tuan and S Shi. (2010). Stem/progenitor cell-mediated de 27. Seo BM, M Miura, S Gronthos, PM Bartold, S Batouli, J novo regeneration of dental pulp with newly deposited Brahim, M Young, PG Robey, CY Wang and S Shi. (2004). continuous layer of dentin in an in vivo model. Tissue Eng Investigation of multipotent postnatal stem cells from hu- Part A 16:605–615. man periodontal ligament. Lancet 364:149–155. 12. Thompson A, J Scha¨fer, K Kuhn, S Kienle, J Schwarz, G 28. Gronthos S, AC Zannettino, SJ Hay, S Shi, SE Graves, A Schmidt, T Neumann, R Johnstone, AK Mohammed and C Kortesidis and PJ Simmons. (2003). Molecular and cellular Hamon. (2003). Tandem mass tags: a novel quantification characterisation of highly purified stromal stem cells de- strategy for comparative analysis of complex protein mix- rived from human bone marrow. J Cell Sci 116:1827–1835. tures by MS/MS. Anal Chem 75:1895–1904. 29. Hilkens P, Y Fanton, W Martens, P Gervois, T Struys, C 13. Gilar M, P Olivova, AE Daly and JC Gebler. (2005). Two- Politis, I Lambrichts and A Bronckaers. (2014). Pro- dimensional separation of peptides using RP-RP-HPLC angiogenic impact of dental stem cells in vitro and in vivo. system with different pH in first and second separation Stem Cell Res 12:778–790. dimensions. J Sep Sci 28:1694–1703. 30. Tvorogov D, A Anisimov, W Zheng, VM Leppa¨nen, T 14. Chevallet M, H Diemer, A Van Dorssealer, C Villiers and T Tammela, S Laurinavicius, W Holnthoner, H Helotera¨,T Rabilloud. (2007). Toward a better analysis of secreted Holopainen, et al. (2010). Effective suppression of vascular proteins: the example of the myeloid cells secretome. network formation by combination of antibodies blocking Proteomics 7:1757–1770. VEGFR ligand binding and receptor dimerization. Cancer 15. Kim JM, J Kim, YH Kim, KT Kim, SH Ryu, TG Lee and Cell 18:630–640. PG Suh. (2013). Comparative secretome analysis of human 31. Carmeliet P and RK Jain. (2011). Molecular mechanisms bone marrow-derived mesenchymal stem cells during os- and clinical applications of angiogenesis. Nature 473: teogenesis. J Cell Physiol 228:216–224. 298–307. 16. Timmers L, SK Lim, IE Hoefer, F Arslan, RC Lai, AA van 32. Hodivala-Dilke K. (2008). avb3 integrin and angiogenesis: Oorschot, MJ Goumans, C Strijder, SK Sze, et al. (2011). a moody integrin in a changing environment. Curr Opin Human mesenchymal stem cell-conditioned medium im- Cell Biol 20:514–519. proves cardiac function following myocardial infarction. 33. Kim I, SO Moon, SK Park, SW Chae and GY Koh. (2001). Stem Cell Res 6:206–214. Angiopoietin-1 reduces VEGF-stimulated leukocyte adhe- 17. Kupcova Skalnikova H. (2013). Proteomic techniques for sion to endothelial cells by reducing ICAM-1, VCAM-1, characterisation of mesenchymal stem cell secretome. and E-selectin expression. Circ Res 89:477–479. Biochimie 95:2196–2211. 34. Sieveking DP and MK Ng. (2009). Cell therapies for 18. Yu B, X Zhang and X Li. (2014). Exosomes derived from therapeutic angiogenesis: back to the bench. Vasc Med 14: mesenchymal stem cells. Int J Mol Sci 15:4142–4157. 153–166. 19. Yeo RW, RC Lai, B Zhang, SS Tan, Y Yin, BJ Teh and SK 35. Kadomatsu K and T Muramastsu. (2004). Midkine and Lim. (2013). Mesenchymal stem cell: an efficient mass pleiotrophin in neural development and cancer. Cancer Lett producer of exosomes for drug delivery. Adv Drug Deliv 204:127–143. Rev 65:336–341. 36. Alguacil LA and G Herrado´n. (2015). Midkine and pleio- 20. Lai RC, F Arslan, MM Lee, NS Sze, A Choo, TS Chen, M trophin in the treatment of neurodegenerative diseases and Salto-Tellez, L Timmers, CN Lee, et al. (2010). drug addiction. Recent Pat CNS Drug Discov 10:28–33. 508 YU ET AL.

37. Zou P, H Muramatsu, T Miyata and T Muramatsu. (2006). 44. Choi YA, J Lim, KM Kim, B Acharya, JY Cho, YC Bae, HI Midkine, a heparin-binding growth factor, is expressed in Shin, SY Kim and EK Park. (2010). Secretome analysis of neural precursor cells and promotes their growth. J Neu- human BMSCs and identification of SMOC1 as an impor- rochem 99:1470–1479. tant ECM protein in osteoblast differentiation. J Proteome 38. Liu JY, X Chen, L Yue, GT Huang and XY Zou. (2015). Res 9:2946–2956. CXC chemokine receptor 4 is expressed paravascularly in 45. Ando Y, K Matsubara, J Ishikawa, M Fujio, R Shohara, H apical papilla and coordinates with stromal cell-derived hibi, M Ueda and A Yamamoto. (2014). Stem cell- factor-1a during transmigration of stem cells from apical conditioned medium accelerates distraction osteogenesis papilla. J Endod S0099-2399:00341-00346. through multiple regenerative mechanisms. Bone 61:82–90. 39. Sloan AJ and AJ Smith. (1999). Stimulation of the dentine- 46. Al-Sharabi N, Y Xue, M Fujio, M Ueda, C Gjerde, K pulp complex of rat incisor teeth by transforming growth Mustafa and I Fristad. (2014). Bone marrow stromal cell factor-beta isoforms 1–3 in vitro. Arch Oral Biol 44:149–156. paracrine factors direct osteo/odontogenic differentiation of 40. Toyono T, M Nakashima, S Kuhara and A Akamine. dental pulp cells. Tissue Eng Part A 20:3063–3072. (1997). Expression of TGF-beta superfamily receptors in dental pulp. J Dent Res 76:1555–1560. 41. Li JY, B Hu, XJ Wang and SL Wang. (2008). Temporal Address correspondence to: and spatial expression of TGF-beta2 in tooth crown de- Dr. Lihong Ge velopment in mouse first lower molar. Eur J Histochem Department of Pediatric Dentistry 52:243–250. Peking University School and Hospital of Stomatology 42. Xu L, C Song, M Ni, F Meng, H Xie and G Li. (2012). Zhongguancun South Avenue 22 Cellular retinol-binding protein 1 (CRBP-1) regulates os- Haidian District teogenenesis and adipogenesis of mesenchymal stem cells Beijing 100081 through inhibiting RXRa-induced b-catenin degradation. China Int J Biochem Cell Biol 44:612–619. 43. Berkovitz BK and M Maden. (1995). The distribution of E-mail: [email protected] cellular retinoic acid-binding protein I (CRABPI) and cel- lular retinol-binding protein I (CRBPI) during molar tooth Received for publication September 21, 2015 development and eruption in the rat. Connect Tissue Res Accepted after revision January 7, 2016 32:191–199. Prepublished on Liebert Instant Online January 8, 2016